Expected behavior
The application insights agent runs smoothly in the background without significant impact on the server or the application.
Actual behavior
In rare cases, one of the "BatchSpanProcessor_WorkerThread-1" thread runs in an endless loop and occupies an entire cpu core.
Stacktrace of this thread:
java.lang.Thread.yield(Native Method)
java.util.concurrent.locks.AbstractQueuedSynchronizer.transferAfterCancelledWait(AbstractQueuedSynchronizer.java:1752)
java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2119)
java.util.concurrent.ArrayBlockingQueue.poll(ArrayBlockingQueue.java:432)
io.opentelemetry.sdk.trace.export.BatchSpanProcessor$Worker.run(BatchSpanProcessor.java:230)
java.lang.Thread.run(Thread.java:829)
When the problem started, some telemetry data was missing in the Application Insight portal.
To Reproduce
The problem cannot be reproduced specifically. It occurred twice in a load and performance test at high load and once in a productive system under normal load.
Sample Application
System information
Please provide the following information:
- SDK Version: 3.2.4
- OS type and version: CentOS Linux 7 (runs in a VM from VMWare), Ubuntu 20.04.4 LTS (runs in a pod on a VM in azure)
- Application Server type and version (if applicable): Tomcat, spring-boot
- Using spring-boot? yes
- Additional relevant libraries (with version, if applicable): -
Logs
No log files, but a heap dump of the Java VM (including the state of all Java objects) is available. For data protection reasons, however, I cannot publish this dump. However, I will be happy to answer any questions about it.
Screenshots
Expected behavior
The application insights agent runs smoothly in the background without significant impact on the server or the application.
Actual behavior
In rare cases, one of the "BatchSpanProcessor_WorkerThread-1" thread runs in an endless loop and occupies an entire cpu core.
Stacktrace of this thread:
java.lang.Thread.yield(Native Method)
java.util.concurrent.locks.AbstractQueuedSynchronizer.transferAfterCancelledWait(AbstractQueuedSynchronizer.java:1752)
java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2119)
java.util.concurrent.ArrayBlockingQueue.poll(ArrayBlockingQueue.java:432)
io.opentelemetry.sdk.trace.export.BatchSpanProcessor$Worker.run(BatchSpanProcessor.java:230)
java.lang.Thread.run(Thread.java:829)
When the problem started, some telemetry data was missing in the Application Insight portal.
To Reproduce
The problem cannot be reproduced specifically. It occurred twice in a load and performance test at high load and once in a productive system under normal load.
Sample Application
System information
Please provide the following information:
Logs
No log files, but a heap dump of the Java VM (including the state of all Java objects) is available. For data protection reasons, however, I cannot publish this dump. However, I will be happy to answer any questions about it.
Screenshots